I’m reading all your ideas and they are very helpful! I’m going to test some the next days as I wire stuff.

I don't know much of music theory though I know someone who does and could ask him if you want, but my programmer brain instantly suggests that you find a way how to import music into a program, then import a bunch of music from different genres and let the program analyze what tones are played simultaneously.

I think there's a ridiculously easy solution (though not always easy to come up with) for you. In RF circuits, we use "mixer" stages to mix 2 frequencies together for many purposes. Now, while in RF circuits we usually filter the outputs to only get one of the multiple "mixed" frequencies, for the piano, you would simply want to "use them all". In this way, if any 2+ keys from one coil's "section" are pressed, any arc that is generated (regardless of which 'hammer' strikes the arc) would generate all the tones for the pressed keys. Important PostScript: High quality "arc speakers" all use a 1MHz (or similar) "main frequency) for the arc; then MODULATE THE CURRENT of the arc to generate the actual audible sound. This gives far superior (by orders of magnitude) sound quality, and would also simplify the problem(s) you're having now by allowing you to mix the multiple "key frequencies" together in an ANALOG manner, rather than having to rely on having the coils & driving circuits try to function similarly across several octaves of frequencies!

I think you should decouple the flybacks from the actual key. In that case you can play 10 notes at the same time and it does not matter which notes that are. So basically you monitor all keys by a microcontroller and when the first key is pressed you assign it to the first flyback and reserve that one until the key is released, when the next key is pressed you assign the next free flyback. It is a bit like a polyphonic synthesizer. The retro/basic models are also limited in voices. I remember my child "piano" could only play two notes at the same time. The other key presses were detected but ignored.

Please don't get yourself killed. You are my favourite KZfaqr (and Engineer).

I remember when you rewired pianos with non-piano strings to kinda see if the strings sounded different. However, after all this time, you are ACTUALLY making a friggin' PLASMA PIANO. Talk about a story arc. It's been quite the journey, and here's to looking forward where you'll go next :D

Hi! I'm currently a music student at university. I'm still learning (please correct me if I'm wrong too), and I'm sure there has been other people who have mentioned some of these ideas, but I wanted to try and help if I could! (P.S. I love your videos!!) For the music theory, minor 9ths are one of the most dissonant intervals that can be played. In theory, that interval isn't played a lot as it comes to our ears of "not sounding good", but that doesn't mean it's not played in music. Mi9ths can be super cool. It might matter on the genre the person may be playing, as for example, jazz can use that interval more often compared to pop. The chords that you should be wary about playing would be playing chords like 7b9. While writing this too, sometimes piano players can hit wrong notes too, so trying to avoid any kind of 7th or 9th chords might be helpful too, so you have a lesser chance of hitting the wrong note. An example of that could be a Ma9 chord. Of course, I'm mainly talking about jazz chords, and these chords are less likely to appear in more pop music, but the chords do sound cool lol. Another area to take into consideration is if the song so happens to be playing a chord, and one part of the melody (or something) plays a note that doesn't so happen to be in the scale tones. (The reason for that being could be for texture, or movement for the song.) For example, if someone would play a C major chord, (just below middle C), and plays a Db (which would be just above middle C), technically, the Db would be a mi2 in the C major, but if you have the chord right below it, it would hit that "mi9" sound as it would be from C2 to Db3. Also, I'm not sure if you'll be playing any scales on this piano (with both hands on top of that haha) and just so happen chance to hit the mi9, there are some scales that include the mi9. An example could be Mixolydian b9 #9 b13. Very cool scale, but it needs to hit the mi2 interval to create that scale. I think mi9 is the best way to go, as a Ma9 would be a lot more likely to hit, and an octave will also be very likely to hit. I think the way I would theorize it would mark every mi9th you are planning on hitting on the piano in a very clear spot, so it's easier to remember, then look up possible songs you are wanting to play and see if there is any collision. Or maybe do the opposite (find the song then mark the mi9.) If you want to set up the mi9 to the flyback first, I would start from middle C and work the opposite way. From my experience, you are more likely to play notes between the 2 octaves down middle C and 2 octaves up middle C. Of course, this matters what you are planning to play, but those notes would probably be more common, and you would have a better control in finding the notes you do want to play, compared to working down from the very left of the piano or the right. (If that makes sense?) I really like some other people's ideas and thoughts in the comments too, such as using a program of some kind to find where a mi9 could be found in the song. I wish I could help there, but I know nothing about programming haha. This was really long, I'm sorry. I hope this made a bit of sense! I got a little too into this music theory haha. But I hope I could help a bit! Good luck!! You got this!! (Also be safe with those electrics!!)

I think that the idea 1 octave +1 is your best option. For 2 reasons: 1) You could of course lay it out harmonically so that keys of a chord are not together on a flyback. Then you can only play songs in the same key. And as soon as you have "exotic" harmonies this won't work anymore. 2) You have to make sure that you don't get in each other's way with the left and right hand. One idea is in the region where you often play with both hands at the same time to use more flybacks and in lower and higher octaves to pack them together. Hope you understand my thoughts :)

Also be careful, very lethal voltages and power levels here. Stay safe please

The octave+1 sounds like a decent solution but that's based on a limited understanding of music theory. Keep up the good work Mattias.

You could analyze a large database of midi songs to see empirically which notes are least likely to be paired together.

Why not one flyback per note played, so if you play one note, one flyback is used. When you play two notes, it uses two flybacks. This is how a analog poly synth works with oscillators. every note has it's own oscillators. So to play 4 notes you need only 4 flybacks, and so on. To be clear, pardon my bad english: One key played uses a flyback, two key played uses two flybacks, and so on. Not assigned to a group of keys, but rather as a analog poly synth does. Only problem is to arc the correct hammer when played. so you need to route the flyback to the correct hammer that is played for the visual effect.

I do have a music degree and I think your octave +1 semitone idea sounds like a really good one if you’re only playing songs where your hands are fairly close together. The more octaves apart you stretch your semitones will start to add up. Also you will need 12 channels to make it work. In the last couple of minutes of thinking I haven’t come up with anything better though.

Well, I’ve read some other ideas, from people who are more on the science side of things, but to answer your question on the music theory side, an octave plus a half step is a good interval idea. Not only is it rarely used (if at all), just musically, due to the mathematical frequency intervals that range would do, that combo would just sound unpleasant. For proof of your idea being good, go to a different piano and actually play a chord that uses that interval, and hear how awkward it sounds. Other complicated solutions that are better (likely) aside, this is a good fallback idea, or if other solutions are tricky or problematic, then this is a good, easy, first idea.

That kind of reminds me of how old clavichords work. One string is in charge of multiple notes. I think they use close together notes for each string, but that might be because of the physics of an actual string rather than your midi tesla coil arcs.

Generally speaking, what a lot of people already wrote: you could connect one key with the key one semitone next to it, then maybe tritonus (6 halfsteps) and then as you did octave + semitone. These should theoretically not occur very often

Awesome idea!! The sound reminds me on thse old, child keyboards from the 90's

The augmented octave is one of the most dissonant intervals, so I think your idea is acha good idea... At least for pop music, your Tesla piano will be very functional.

from my response to another comment, because I could code this in a few minutes: You have ten memory slots, and when you depress the first key, the note goes in slot 1. When you press the second key, that goes in slot 2. When you release the first key, that note is taken out of slot 1. When you depress a third key while still holding the second, that note goes in slot 1. Each slot corresponds to a flyback. Once ten keys are depressed, no more slots are available to hold notes, so pressing an 11th key is ignored.

In my opinion the best way to handle this issue is to use digital instruments and map the keys while someone is playing them this will give you the likelihood of what are played together because I believe that you are 100% correct about splitting the circuits between the flybacks in a staggered pattern so if you mapped the keys from someone playing a keyboard you would be able to draw the probability of keys played together and then use that as an algorithm to plan which key is controlled by which flyback

Do you want one arc per key struck or just an arc regardless because remember, the arc is just a speaker and can reproduce multiple notes, ie if you only have one coil in a location where you're playing two keys, you can just generate a single arc but still reproduce both notes in the waveform..... Alternately, if you just want as many arcs as possible, you might just alternate each key with a different coil (ie interleave) and basically decouple the special effect (the arc) from the sound generation (because *any* arc can generate *all notes*), so your program logic is simple to turn on whatever coils correspond to depressed keys, but you just send the same combined sound signal to all coils just like a "master" speaker vs having one speaker per note...

In my mind this solution is a bit too complex. You create one tone only with one flyback, but a radio has only one speaker, and still able to recreate endless harmonies. I guess what I say is that the creation of the tones should not be done one by one, but the controller should mix all the keys that are depressed, and send that premixed output to all the coils. I saw someone said the RF radio works somewhat the same, and since I tinker with many older superhet tube radios I believe that this is an easier way indeed. This way you only need the many coils just to ensure that all hammers that are pushed would start a discharge, and not just the one that was deppressed first, but all the plasma discharges would play the same pre mixed notes. I think at least... I love your videos, cheers!

This already looks epic, I hope you can get it working.

you can basically generate square waves with low duty cycle for every key that is pressed, using a microcontroller or something, and you can just OR the signals together and send that to the flyback driver.

This is precisely why we used qwerty keyboard so the hammers didn't stick on the typewriter (commonly used letters together further away)!

I love how you take such a simple and common instrument like a piano and make it so dangerous and scientific.

I’m a mathematician, and I just need more music theory knowledge to be able to answer this. But there’s probably a fancy solution

Making very high and especially low notes into the same thing (so low notes close to eachoter as one) because playing two low notes one step apart together sound very muddy and is rarely used.

I think this is a great solution, it’s the one I thought of too.

with all of the useful suggestions, watch Mattias make better plasma piano than actual pianos

You could make the first monophonic piano (I pissed my self laughing when you started playing). This it the most in tune piano of yours I have heard!

What is very rare is that you play the highest octave and the lowest octave together, so you can wire them up on the same circuit. I'm an intermediate piano player and I use the following 2 neighbouring octaves 1 in between 2 in between 3 in between but that's usually only a moment, so you might be able to ignore that. It limits the advanced songs you can play. But enough for this prototype to look really cool.

A octave has 12 halve tone so if you do the octave + 1 method you would need 13 transformers. That's more than you want. If you do octave -1 you have the same effect but with only 11 transformers.

Can the Arduino merge frequencies together? If you're using MIDI you can look for digital solutions. Others already mentioned synthesizers, this is the polyphony problem they've been solving forever.

Instead of putting one flyback for the same octave, I suggest placing one flyback per note on octave, same flyback across octaves. Of course it will be necessary to wire the lower octaves with a different flyback. that would mean 14 flyback total and you could play any piece of song, as long it does not go in the same clave (is rare for that to happen across different clave and it would need to be the same ton one octave lower). This is based on the fact that humans cannot play more than 8 notes with their hand span. the problem is that humans have two hands, for one hand play this solution would work. However, after see other answers, I could suggest using something like a multiplexer, so you switch what banks of notes the flybacks are wired to. How to do that would be kinda crazy... now the problem is just pure electronics topology and engineering.

This is so cool

My brain is too small for this

Sneak into your neaighbods house and borrow their electricity and mix it with yours to save money

True engineer

Can't you play a chord through the flyback driver, by having 2 different frequencies mixed together ?

Use a diesel generator as power source. So you wouldn’t need to compromise the performance of the piano and keep your power grid alive.

To make it work with decoupled transformers it would probably have to work like this... when a key gets pressed the signal gets sent to the micro controller, it will turn on a relay for that hammer and then assign the first flyback to it, then the same with the next key, the only issue with this is that you would need a separate relay for every key times every flyback. So ten relays per key, witch will be very hard to make/be too expenvive. But.... if you put one relay on each key, going directly to the flyback (main power not signal) then you could use a differnt flyback on each key, because they will only be powered when a key is pressed. Just make sure to have a breaker between the outlets and the main circuitry so if it surges from pressing too many keys ut stops before it even gets to your outlets. It solves your power problem unless you press more then 10 keys at once. Also using a relay system will remove the need for a micro controller monitoring each key and will only be needed to generate the frequencies. Good luck and we love you Mattias!

You are making the world’s most dangerous synthesizer. I also think like wait why can’t you have a 3KW piano?! 😂. In all seriousness, I am very excited for this piano video. I think your idea about using an octave isn’t a bad one, but chords and melody at the same time? I feel like the classical pianists you get are gonna feel weird about playing it.

Perhaps one way to do this would be to get a bunch of piano sheet music, and build some sort of algorithm that reads and maps out keys commonly played in association. Might have to stick to one genre. Maybe you could have ten flybacks wired to all your keys and then dynamically allocate them?

I think you could add more flybacks without too much issue as unless I'm mistaken they only really draw power while arcing. Worst case scenario if you play a track that's too fast you pop a breaker assuming your house is up to code. You might not be able to do one for every key but you can reduce the overlap a fair bit more

I am not sure how the circuits generate the frequency that goes thru the flybacks, but a physics solution would be: you could generate a signal which is the superposition of all of the frequencies being played, and send it thru the flyback, the thing with this method is that all signals will go out from one spark only.

Oh my god it's a QWERTY piano problem

That's a really great idea, if you want a cheat - put it on two notes that are right next to each other like C and C# (not specific) The reason for that is that two notes that are only half step apart sound really dissonant when played together so it's rarely used in things like jazz and even then I don't think it occurs much

Only potential Achilles heel for the octave +1 method is if you start using 7 chords (C, E, G,B). However, these chord would have to be built in such a way that the root and 7 of the chord are at that octave +1 point. Just solely from a theory perspective.

I'm not sure how to solve the problem but I'm sure that the wood that transformers is on should be replaced with something Fire Resistant

This is so amazing!! I had no idea your insane sounding project became something so... Well, musical :) If you have some programming friends, I think AI would be a good bet. Feed a bunch of piano midis into an algorithm and find the notes that are almost never played simultaneously. Though, maybe that is easier said than done!

So if I understood the problem correctly you have 10 flybacks and they need to be in the correct position before the player pressed the key If you know what songs get played you could use the midi transcription to put them in the correct position before the key press For improvisation it seems a bit harder though You have two hands with five fingers each and 10 flybacks so you can dedicate five to each hand Each hand has just a span of notes that it can physically play at once If you just look at the left hand you maybe can just update it so that the span gets moved down to the bottom position of the pinky finger for this hand (= the lowest note getting played) Then you will still need some criterium to split the hands though Id say the span of maybe a 9nth chord would be the maximum you need to worry about for simultaneous notes

This will work better with more flybacks but when two keys sharing the same flyback are pressed you could output a waveform of both sounds combined That way it's always playable in terms of audio and visually you get arcing on every key individually but when keys belonging to the same flyback are played simultaneously you only get one visal arc Not a perfect solution but its a good starting point

It might be most effective to limit the amount of scales the keyboard works in. More often than not, most groups of a random three notes are going to have an assortment in a key or scale they fit into. Due to the tuning of older harpsichords, they just didn’t sound good in certain keys. The plasma piano could be a spiritual successor to that in some ways lol

I love your Hammer Piano VST BUT... if you want to make one that plays easier or even experiment with different materials you can add pins or anything to the original hammers (metal pins is the professional way of making piano sound sharper I believe) instead of replacing entire hammers, though the idea was hilarious.

I wonder if GhostBusters' thing with never crossing the arcs is actual here too

it just teleported into my hand...

You have a fully functional tesla speakers. Which means that the best way would be to separate the tesla action from piano. Use piano as midi input to synthesiser. Synthesize large groups like octaves into sound, maybe virtual hammer piano? And then output it to the tesla coils...

I believe your solution is a good one. However I got something for you to keep in mind when using this solution: Be careful with the +1 idea when you get higher on the piano. So you don't make a common chord like C and E (+4) together. Make it like +3 instead or something like that for that specific octave. And as always remember to keep a note near by so you don't lose track of what you're changing. Wish ya the best and good luck! (from) Everyone sus rett

For me it seems like some how multiplexing the flybacks will do the job

The b9 interval make sense when you think about it from the root note. Unless you are a jazz player and play lots of b9 chords or course. BUT... Where it will go wrong is when you start stacking notes on the chords. Ie if you played 1,3, 5, 11. The 11th wouldn't sound presumably because the interval between the 3rd and the 11th is a b9. ie in C this is CEGF And this is probably way more common than you think at first because you are just thinking of the b9 from the root and not the middle of the chord. And of course the 11 interval is probably more of a major tone than any other because it the highest note and more likely to be part of a melody which is another problem, you want it to sacrifice the lower note not the higher.. You could also come into this problem with say playing minor + b6 in melody... Eg ACE..F. or maj7 with root in melody... Ie CEGB...C. I am not a pianist but this was just off the top of my head... I think you'd be surprised how much this would affect melodies if it always prioritised the lower note. I liked the idea someone else had where you are probably less likely notes within the octaves furthest apart on the paino the. Maybe you could also plus/minus 1 semitone...but not sure if that's mathematically practical with what you are trying to do. Ie Most polyphonic in the middle and monophonic at the edges...I would imagine the most harmony clusters are in the middle of the piano. Like you would hardly ever play a C1 and D1 together. But a C4 and D4 yes. There isn't really a set interval that would work across the whole range of the piano.

*I am not an engineer* or a programmer, so reading all these comments hurt my brain! 😉

maybe if you want a plasma to show up when playing a chord, maybe you can use 1 circuit for maybe like 5 keys. Maybe that'll work

Also never wear gloves when dealing with high voltage. For AC it does not work and for DC it can make it far worse. Since your hands works as capacitors, meaning your body will build up a potential till you accidentally discharge that current with another body part and get a proper electrical shock. Professionals uses faradays cages. Metal conductors connected to ground. So when you accidentally touch high voltage stuff it takes the easiest path to ground, which is the metal that protects you and is connected with the ground. Cars protect you from lightning for the same reason.

In theory, you can go "poor speaker" all the way and play multiple frequencies on a single flyback at once. That just means a bit more math on the arduino side, but it's technically possible to handle everything with only one flyback. (but that might feel like cheating ^^')

Well, not having a resistor in series with the flybacks' secondary is not the smartest idea. You might cook the transformers.

Did you ever finish this? It looks amazing

I would cycle it in tritones. It's an interval that's normally avoided but than octaves should be a problem. You can avoid it tho going tritone, tritone+½ (octave +1 note)

Depends on which music you want to play. Genarally the deeper the tones the wider the distance between them, the higher the more narrow. Look at the overtone scale to get an idea! Seldom are for instance a chord like c-e-g-c on the very left of the piano. c-g-e is more usal, or just an octave c-c. In high areas is c-g-e däremot somewhat unusal. This results in that many pianists (including me) have wider left hand. But be aware there pieces som are exceptions!

someone right now is download sheet music for the last 100 years and writing a program to find any and all key combos played on a panio to find the best key placement. Its likely to give you a 95% sucess rate of never overlapping while there is those few songs to where you cross hands or comes right close. I have no skill other than my basic knowlage and understanding of things. Though I do think that ia uni student working on ai or some scripting or something could indeed build out a program from an online souce tha shows sheet music by the notes. Would not take more than a day of processing with a pretty high end computer and few to 6 hours to write the code. My general idea is to assign a number to each key to represet the sheet music. Then assign the proper finger and hand to those notes. Then a bunch of other true false things. Since you want ot find out what set of keys are never played at the same time/right after another ti will compile out a list and then using said list should be able to break it down in to 10 segments to he correct 10 keys. As you may find that key 1 and key 16 are never played togeather 100% of the time out of the 50k songs where key 1 and 10 are played

what if someone plays with 2 hands? :< i heard that happens sometimes

You need one flyback per key, but then you can implement a polyphony limit at the software level.

Just use 1 flyback and ply multiple tones through it. That's how real electric pianos get away with 1 speaker anyway. This is done using a 16khz or higher pwm to send the sound samples through.

Well it all depends on what you are planning on playing. If you want to keep it simple choose a key and then anything outside of that key don't play. For example the key if C uses only the white notes so that would allow you to do anything within the key without using any black notes. Another option is to think about pentatonic scales. Essentially find music that falls within the pentatonic scale which limits you to only using the1,2,3,5, and 6th notes in any key. If I were personally working in this project I would probably just use the key of C and make sure to tell anyone that wants to play it that the key of C is the only one that will work. That way you can have a flyback work with just an octave in that key. Then assign them to each octave. In the end yes you are limiting the potential of playing sharps and flats and using different keys. But any competent musician should be able to easily transpose to C and any one who is not a master at music would have absolutely no idea because in the end it all still sounds great.

Seems like you need to electrically isolate each hammer/plate so that consecutive keys can be played without crosstalk.

I don’t know much about the electronic stuff, but no, it’s not super common to play a minor 9th (octave +1), it is pretty common to play the 7th however. The difficulty you’ll have is left hand and right hand being close together, though, I’m sure there are songs which keep them apart. Also, a talented pianist could probably adjust to any technical issues that you give them. So if you tell them not to play any 9ths and they could probably adapt to that, etc.

If you could somehow set it up so that one flyback plays the first note on any of the 88 keys, if only one key is pressed. Somehow you could isolate that circuit, so a second flyback would activate and play the second note from one of the remaining 87 keys if played simultaneously, or sequentially if the first note is sustained, and isolates itself from the remaining 86 keys. If a third note is played, that Arduino/flyback/hammer/plasma circuit would isolate itself from the remaining 85, etc... And as long as you're not playing any songs that call for four hands(20 fingers), you should be ok with 10 circuits at a time.

I challenge the idea that you can't have more than 10 because they use too much power. Because they won't all be turned on at the same time so why will they be using power? I guess you will have a switch underneath each key because you need that in your plan to select the frequency anyway, so why not also use that to turn the flyback on when needed?

use violin strings and then play worlds smallest violin as a meme

Is there a way (it would be a lot of wiring) you could connect every flyback to every key and switch it out with a big switching board so that each key played with each of your ten fingers gets its own flyback?

If your controlling with an arduino then you can just have it detect when you press more than one key with a flyback, then only power one depending on which would be the best way to do that. This could be a very simple algorithm or something more complex to do interesting stuff.

how many volts and amperage are generated by one Flyback when key is pressed? for one flyback connecting each key to separate relay or contactor could rewire hammer connections to that given flyback in theory, as I'm not sure how would handle relay or contactor so much power. If it would be low like 240v and up to10A than cheap relay shields for arduino would help like 10$ for shield containing 8 relays (or better 16 channel relay shield), 11 those shields would be required to handle one flyback and whole keys, so if you want press more than 1 key (like aiming the max 10 of fingers) it would require 10 flybacks and 110 shields with total 880 relays on them, and as well some demultiplexers for controlling all of those relays by dialing on small amount of arduino pins a specific address (binary) of what output demultiplexer pin will be 'turned' on. But I can only imagine its more than 240v and there should be verified if there would be circuit power jumps over the board. You told that 10 need would generate 30k volts it looks like 340v per flyback but how many amps? Higher voltage can handle contactors as they are dedicated for heavy liftings cost of it would rise bit more I guess like expensive 22$ handling 300v 12A 100$ would be with parameters handling 440V 25A and controlled by signal 12V DC, 1000$ dollars handling 240V 320A that weights 10kg and you would need 880 of them, but you would have to consult it with someone who knows more I might be wrong at any of above statements.

I’m wondering if someone has looked at what the most played notes/keys are and make sure that the top ones are on separate circuits

You need a polyphonic tone for the flyback

I think you can play multiple tones at the same time through one flyback. just like you can play multiple tones through one speaker. If you do this you can't use the arduino nano though.

Use a impar ratio like old gearboxs for cars. Also include the 12 tones. Obviously you want each octave to be on its own set especially close octaves.

Have you tested if the arcs of two adjacent hammers fuse mid air? As the distance between the arcs is lower than the distance to the back. That would distort the sound wouldn't it?

What kind of circuit are you using to drive the flyback transformers currently? There's not really any reason that flybacks can't play multiple notes at once, the flyback and arc in air acts as a low pass filter for audio, so with an appropriate driver it can function as a class d amplifier.

If you have MIDIs of all the songs you want to play I can write a script that calculates which notes are never played together.

I might be on the wrong track, but I feel like you should look up how old FM synth keyboards did their polyphony matrix. They had similar limits on how only a few keys could be played at once.

Use aCapasitor to store 30,000 volts

I am wondering whether a scan matrix would work for this. Given you already have a switching transistor in your circuit you just need to expand this out for each key and then in software limit which keys key get energised and limit it to the max power your PSU can supply. You would need to have one transformer per key though. It's a technique we use in touch sense of motorised faders to energise the person touching the fader.

The problem with moving up an octave +1 key is that it leaves 11 other keys between. If you're using just 10 arcing "registers" you won't have enough. If you space them 10 half steps apart, that MIGHT be enough spacing for most music, but could have some problems. I could imagine a spacing which is different in the high notes where songs often play, and is more clustered among the low notes, but that'd be complicated. The issue with finding a perfect solution is that different genres of music tend to have different intervals be common. Classical music abhors tritones, but jazz loves it. Rag time will only stride on the low notes, but classical will use lots of it at once. So if you want something perfect, you go with the microcontroller approach suggested elsewhere. If you want something simple, try tritones spacing if you mostly play classical-esque music.

Would there be a way to have the Flybacks assigned on-the-fly to the different keys? Are you creating the high voltage only when the tone is played, or is it constantly there and by the movement of the hammer, the arc is created?

What if you were to “add up” the sounds if you press more than one key and have the fly back play that frequency instead of a single note. Like the plasma speakers where they play the frequency of the sound and not of individual notes.

Have you checked what the wiring in your house/apartment is rated for in the kitchen? In the US we have higher rated systems in place for kitchens to ensure they can handle all of the different appliances. Additionally, the same can be said for the clothes washer/ dryer outlet. That is my first thought for an easy solution (work smarter not harder). Your other thought is also what I was thinking for the next easiest solution, but I only have slightly more music experience than you so I have no idea what would ensure you never hit keys on the same wiring section at once. It would make the wiring look pretty ugly and a bit harder to follow so make sure you take extra time on it to make your life easier if something needs to be fixed or changed.

Not entirely sure if this solution is practicle but could you connect all the power sources up to all the keys with an electronic switch between. When multiple keys are pressed the keys are assigned a power source and the switch between that source and the key is flipped.

Woop, clicked so fast

Can the hardware generate more complex wave forms than like a 1-frequency sine wave? I don’t know much about the hardware, but if this were me I’d just be generating waves in software that combine the multiple frequencies-(and that shouldn’t be too hard, either inverse Fourier transform some Dirac deltas or just use someone’s pre-existing library.)

if you play a c with a c# you get dissonance, kinda cool in the right place, i think i would try to use a diminished scale, give me feed back so i can help you further more, is several take a option ?